Skip to content

Logistic regression, Lasso, SVM, Decision Trees, Random Forests

Notifications You must be signed in to change notification settings

MartaFatto/The-Hit-Song-Predictor

Repository files navigation

The Hit Song Predictor

Overview

Our work is inspired by the so called Hit Song Science, whose pioneer is the music entrepreneur Mike McCready. The Hit Song Science aims at predicting whether a song will be a hit before its distribution, by analyzing its audio features through machine learning algorithms.

Research question: "Can we accurately predict whether a song will be a hit knowing some of its audio features?"

Statistical Learning and Machine Learning techniques for classifications:

  • Logistic Regression, Lasso Logistic Regression
  • Support Vector Machines
  • Decision trees
  • Random Forests

Dataset

https://www.kaggle.com/theoverman/the-spotify-hit-predictor-dataset

Required R libraries

The Code is written in R 4.0.3.

  • tree 1.0.40
  • ISLR 1.2
  • ggpubr 0.4.0
  • ggplot2 3.3.3
  • glmnet 4.1
  • MASS 7.3.53
  • randomForest 4.6.14
  • e1071 1.7.4
  • gbm 2.1.8
  • caret 6.0.86
  • dplyr 1.0.3
  • reshape2 1.4.4
  • scales 1.1.1
  • pheatmap 1.0.12

Authors

Licence

This work is available under the Creative Commons Attribution-ShareAlike License. Read more about this license from https://creativecommons.org/licenses/by-sa/3.0/.